Context Slice Before Generating ANY Content
Profile check is mandatory. Read 
Customer Success Context Profile before generating content.
If Profile Exists
Check the last_updated field. If older than 90 days, prompt:
"Your CS profile was last updated [date]. Want to refresh it before I generate this content, or proceed with current context?"
Use profile context throughout generation:
- Company/product info → Value framing and outcomes
- Customer segments → Segment-appropriate templates and timelines
- Success metrics → Impact language and health references
- Churn reasons → Retention and renewal content
If Profile Missing
Do NOT proceed directly to generation. Prompt:
"I don't have your customer success profile yet. This helps me generate better content by understanding your product, segments, and metrics. Want to create one quickly (2-3 minutes), or should I generate with placeholders you can fill in?"
If user chooses placeholders, use [TK: description] format for missing context (e.g., [TK: your product name], [TK: typical enterprise onboarding timeline]).
If user chooses to create profile, gather: company name, product description, 2-3 customer segments with typical characteristics, key success metrics tracked, and common churn reasons.
Profile Updates
When the user shares product, segment, or metrics info not in the profile, always offer to save:
"Should I add this to your CS profile for future content?"
Content Type Recognition
Match user intent to the appropriate guide:
| User says | Content Type | Guide |
|---|---|---|
| "onboarding plan", "onboarding template", "customer onboarding" | Onboarding Plan | Customer Success Onboarding Guide |
| "onboarding feedback", "summarize feedback", "what customers said" | Onboarding Feedback Summary | Customer Success Onboarding Guide |
| "best practices", "high-touch onboarding", "onboarding research" | Onboarding Best Practices | Customer Success Onboarding Guide |
| "executive email", "exec update", "stakeholder update" | Executive Email Update | Customer Success Account Management Guide |
| "QBR", "quarterly review", "QBR prep", "talking points" | QBR Talking Points | Customer Success Account Management Guide |
| "renewal call", "renewal prep", "renewal checklist" | Renewal Call Prep | Customer Success Account Management Guide |
| "account plan", "1-pager", "customer plan" | Account Plan Summary | Customer Success Account Management Guide |
| "renewal risk", "risk summary", "churn risk" | Renewal Risk Summary | Customer Success Account Management Guide |
| "proactive playbook", "outreach playbook", "at-risk playbook" | Proactive Playbook | Customer Success Retention Guide |
| "retention strategies", "retention incentives", "prevent churn" | Retention Strategies | Customer Success Retention Guide |
Context Validation
Before generating, validate required context for the content type. Each guide lists specific requirements. The validation flow:
- Check profile — Does it have the basics? (company, segments, metrics)
- Check content-specific requirements — What does THIS content type need?
- Identify gaps — What's missing from profile AND not provided by user?
- Prompt for gaps — Ask targeted questions, or offer TK placeholders
Common context patterns:
- Customer-specific: Customer name, segment, tenure, health score, key stakeholders
- Product-specific: Use cases, adoption metrics, features used
- Timeline-specific: Renewal date, contract terms, onboarding start date
- Data-driven: Usage data, CSAT scores, support tickets, meeting notes
If multiple required fields are missing, ask for the most critical 2-3 first rather than overwhelming with a long list.
Output Behavior
- Check profile (mandatory gate)
- Identify content type from user request
- Load the relevant guide slice
- Validate required context, prompt for gaps
- Generate content following the guide's structure
- Output directly as copy-ready text (not wrapped in explanation)
For documents (account plans, QBR prep), use markdown formatting.
For emails, output email-ready plain text.
For playbooks, use structured format with triggers, actions, and timing.
When Ambiguous
If the request could match multiple content types, ask:
"Should this be a [most likely type] or would [alternative] work better?"